Timing of integration into the chromosome is critical for the fitness of an integrative and conjugative element and its bacterial host

Integrative and conjugative elements (ICEs) are major contributors to genome plasticity in bacteria. ICEs reside integrated in the chromosome of a host bacterium and are passively propagated during chromosome replication and cell division. When activated, ICEs excise from the chromosome and may be transferred through the ICE-encoded conjugation machinery into a recipient cell. Integration into the chromosome of the new host generates a stable transconjugant. Although integration into the chromosome of a new host is critical for the stable acquisition of ICEs, few studies have directly investigated the molecular events that occur in recipient cells during generation of a stable transconjugant. We found that integration of ICEBs1, an ICE of Bacillus subtilis, occurred several generations after initial transfer to a new host. Premature integration in new hosts led to cell death and hence decreased fitness of the ICE and transconjugants. Host lethality due to premature integration was caused by rolling circle replication that initiated in the integrated ICEBs1 and extended into the host chromosome, resulting in catastrophic genome instability. Our results demonstrate that the timing of integration of an ICE is linked to cessation of autonomous replication of the ICE, and that perturbing this linkage leads to a decrease in ICE and host fitness due to a loss of viability of transconjugants. Linking integration to cessation of autonomous replication appears to be a conserved regulatory scheme for mobile genetic elements that both replicate and integrate into the chromosome of their host.


Introduction
Horizontal gene transfer (HGT) is a major driver of prokaryotic evolution, enabling the movement of genes from a donor organism to a recipient and subsequent maintenance of horizontally acquired genes within a population. Genes transferred horizontally can confer a wide range of traits, including antibiotic resistances, metabolic capacities, pathogenicity, and symbiosis (reviewed in [1]). Conjugation, a form of HGT, is the contact-dependent transfer of DNA from a donor to a recipient cell. Transfer is through a conjugation machine (a type 4 secretion system, T4SS) that is typically encoded by a conjugative element in a donor cell [2][3][4].
There are two broad classes of conjugative elements. Integrative and conjugative elements (ICEs, also known as conjugative transposons) normally reside integrated in a host genome and can become activated to excise and express the conjugation machinery [1,5]. Conjugative plasmids are extra-chromosomal and also contain genes needed for their transfer to recipient cells (1)(2)(3)(4)6). Conjugative transfer of ICEs and plasmids originates at the origin of transfer (oriT) in the element and is initiated by the element-encoded relaxase protein that nicks and becomes covalently attached to the 5'-end of the nicked strand. The nicked DNA is unwound and linear single-stranded DNA (ssDNA) with the attached relaxase can be transferred from the donor to a recipient through the conjugation machinery [6]. Once inside the recipient, the ssDNA circularizes, and undergoes second-strand synthesis to generate circular dsDNA. ICE genes can now be expressed and the double-stranded circular form of the ICE is a substrate for integration into the host chromosome [1,5].
Unlike that for conjugative plasmids, stable acquisition of an ICE requires integration into the host chromosome where the ICE is then passively propagated as the host replicates and segregates its DNA to daughter cells. Many ICEs are known to undergo limited rounds of autonomous replication and both replication and integration are often essential for stable acquisition of the element [1,[7][8][9][10][11][12][13][14][15][16]. The requirement for integration for stable acquisition of an ICE is a clear delineation between conjugative plasmids and ICEs and introduces an area for differential selection and evolution of regulatory schemes.
Little is known about the regulatory factors that drive integration and stable acquisition of an ICE after transfer to a new host. Analysis of events in transconjugants has typically relied on end-point readouts of overall mating efficiency to discern the importance of ICE functions, thus obscuring the context in which a particular function is important. A primary challenge for these types of studies is that transfer events are relatively infrequent and, for most ICEs, only a subpopulation of cells participate.
We used ICEBs1 (Fig 1) from Bacillus subtilis to investigate the timing of successful integration and acquisition in transconjugants. ICEBs1 is well-suited for investigating conserved ICE dynamics as it has been well-studied, is easily manipulated, and can be activated in a relatively large fraction (~20%-90%) of cells in the population [17][18][19] ICEBs1 integrates via site-specific recombination into one site in the B. subtilis chromosome: in trnS-leu2 which encodes a leucyl-tRNA [17,19]. The specific integration site (attB; attachment site in the bacterium) allows for precise measurements of integration (stable acquisition).
Using a combination of population-based and single-cell assays, we found that integration in transconjugants occurs several generations after transfer of the element. Deletion of xis (encoding the recombination directionality factor that is needed for element excision from the chromosome) caused premature integration in transconjugants and was lethal for most transconjugants. Lethality was caused by integration of an active element that was undergoing autonomous rolling circle replication. Our results highlight the importance of ICE regulation in transconjugants and reveal the important underlying linkage between the cessation of autonomous replication (a plasmid-like function) and integration.

Timing of ICEBs1 transfer and integration in transconjugants
Integration of an ICE into the genome of a new host is essential for the stable acquisition of the element. Following the transfer of ICEBs1 from a donor to recipient, the element is capable of rapidly transferring through a chain of cells [20], indicating that the element is active for a period of time before it integrates into the chromosome. We sought to determine when ICEBs1 integrates into the chromosome of transconjugants.
Strains for distinguishing integration in donors and transconjugants. ICEBs1 has a unique integration site (attB) in the bacterial chromosome. When unoccupied, attB and the flanking sequences are identical in both donor and recipient cells. This makes it virtually impossible to physically distinguish, at the molecular level, integration or excision in transconjugants from that in the initial donor. In order to detect integration in transconjugants separately from the initial donors, we constructed a donor strain (SAM078) in which the site of ICEBs1 integration was in amyE (a nonessential gene often used for inserting cloned genes) rather than its normal position in trnS-leu2. We could then distinguish integrants in the transconjugant population (integration in the normal chromosomal site) from those in the initial donor (at amyE, the ectopic location) using PCR primer pairs that were specific for the sequences flanking attB in transconjugants ( Fig 1B). We normalized the number of integrants in the transconjugant population relative to a genomic marker (ydds::mls) that was present in recipients (and transconjugants), but not in donors. yddS is located~5 kb downstream of attB.
Timing of acquisition of ICEBs1. We measured the conjugation efficiency at various times after the start of mating (Methods). Transconjugants were determined by the number of colony forming units (CFUs) with indicative antibiotic resistances (Methods). Recipients (SAM103) lacked ICEBs1 (ICEBs1 0 ). ICEBs1 in donors (SAM078) was activated by overexpression of rapI from an inducible promoter at an ectopic locus (Methods). RapI causes the ICEBs1-encoded protease ImmA to cleave the element repressor ImmR, thereby causing derepression of ICEBs1 ( Fig 1C) [17,18,21,22].
There were typically 5-10% transconjugants per initial donor after two hours and approximately 30% transconjugants per donor after three hours of mating (Table 1). The detection of transconjugants as CFUs at any time post-mating indicated that ICEBs1 had transferred from a donor to a recipient, but did not indicate when ICEBs1 had integrated into the chromosome of the transconjugant.
Timing of ICEBs1 integration in transconjugants. We measured the fraction of transconjugants in which ICEBs1 had integrated at two and three hours after the start of mating. Briefly, genomic DNA was isolated from cells that were recovered from mating filters and we used qPCR to measure integration of ICEBs1. Because the location of the integration site was different in donors (SAM078) and recipients (SAM103), we could measure integration of ICEBs1 in the transconjugants (recipients) using a primer pair that would detect the junction between the right end of ICEBs1 and the host chromosome (attR) that is created only if ICEBs1 had integrated into the chromosome of a recipient. We also determined the total number of recipients in the population by qPCR of a sequence (mls) present in recipients (and transconjugants), but not in donors.
We found that ICEBs1 had integrated in approximately 5% of the transconjugants after two hours and approximately 70% after three hours of mating (Fig 2A). Based on these findings, we conclude that there was a delay between the time of ICEBs1 transfer and integration and that two hours post-mating, ICEBs1 had integrated into the chromosome of a relatively small fraction (~5%) of transconjugants. By three hours post-mating, ICEBs1 had integrated into the chromosome in the majority (~70%) of transconjugants.

Loss of xis caused premature integration in transconjugants
In many integrative elements, including temperate phages and ICEs, recombination into and out of the host chromosome is catalyzed by a recombinase, often named Int (integrase). For many elements, a recombination directionality factor, often called Xis or Rdf (excisionase or recombination directionality factor) is needed, along with Int, for recombination (excision) out of the genome. ICEBs1 encodes both Int and Xis; Int is needed for both integration and excision, and Xis is needed just for excision [19]. Regulation of ICEBs1 is such that int is expressed at a low level constitutively, and xis is expressed at a high level immediately upon de-repression of ICEBs1 gene expression (Fig 1A and 1C) [19].
We found that loss of xis caused early integration in transconjugants. We used donor cells that contained ICEBs1 without xis (ICEBs1 Δxis; SAM207). To enable excision, we provided xis in trans from an ectopic locus, essentially as described previously [19]. Thus, ICEBs1 Δxis could excise in this donor, but there would be no xis in transconjugants. At two hours postmating, ICEBs1 Δxis had integrated in~50% of transconjugants, an increase of~10-fold

PLOS GENETICS
Timing of integration of a mobile genetic element is critical for fitness of the element and its host  relative to that of wild-type ICEBs1 (xis+) (Fig 2A). Loss of xis also caused a decrease in the number of viable transconjugants per donor. At two hours post-mating, there were~0.6% transconjugants per initial donor compared to~10% for wild type (xis+) ( Table 1). Early integration of ICEBs1 Δxis was due to the absence of xis in transconjugants and not loss of a site or an effect on a downstream gene. We expressed xis from its own promoter (Pxis-xis, integrated as a single copy in the chromosome) in recipients. When these recipients (SAM379) were used in a mating with ICEBs1 Δxis donors (SAM207), the timing of integration was similar to that of wild-type ICEBs1 (xis+) (Fig 2A). Additionally, conjugation efficiency in the recipient expressing xis increased relative to that of ICEBs1 Δxis (Table 1). These results indicate that early integration of the Δxis mutant in transconjugants was due to loss of xis and not an unexpected secondary effect on downstream genes or loss of a site. We conclude that xis is needed in transconjugants for proper timing of integration.

Early integration caused cell death
Our data indicated that many of the initial transconjugants that acquired ICEBs1 Δxis were likely dying. For example, three hours post-mating, the integration efficiency per viable transconjugant of ICEBs1 Δxis was greater than 100% (Fig 2A). This indicated that ICEBs1 Δxis was integrating in the transconjugants, but that many of these transconjugants were unable to form colonies and that the resulting increase in the ratio of integrants per transconjugant CFU was likely due to the decrease in transconjugant CFUs. To further investigate this, we monitored the fate of individual transconjugants using fluorescence microscopy.
To distinguish donors, recipients, and transconjugants, the three cell types present in a mating mix, we constructed recipients that constitutively produced red fluorescent protein from mApple and donors that constitutively produced yellow-green fluorescent protein from mNeongreen that had been introduced into ICEBs1 (ICEBs1 mNeongreen). Transconjugants produce both mApple and mNeongreen and were easily distinguished from both donors and recipients that expressed only one fluorescent protein.
We found that a higher proportion of transconjugants that had acquired ICEBs1 Δxis lysed compared to those that had acquired ICEBs1 xis+. In these experiments, we mated donors (xis+, SAM288; Δxis, SAM472) with recipients (SAM271) for four hours and spotted the mating mixture onto agarose pads to observe individual cells by time-lapse microscopy for three hours (Methods). Transconjugants were identified as cells that had both red (mApple) and yellow (mNeongreen) fluorescence. During the three hours of observation,~7% (51 of 732 cells observed) of Δxis transconjugants lysed, compared to less than 0.2% (1 of 635 cells observed) of wild-type (xis + ) transconjugants ( Fig 2B and S1 Movie). We suspect that Δxis transconjugants continued to lose viability after the three hours of observation. per recipient. Acquisition efficiency was measured by qPCR by amplifying across the inversion sequence (which would indicate Cre-mediated recombination) and normalizing to a unique sequence (mls) in recipients. Donors: WT (SAM830); Δxis donor (SAM892). Recipients: WT (SAM599). Circles indicate the results from independent experiments. Error bars indicate standard error of the mean. Results are from the averages of at least two dilutions within the linear range with three technical replicates each for CT values from PCR, then normalized to CFUs. D) Same experiment as in Fig  2C. qPCR from amplification across inversion sequence normalized to transconjugant CFUs. These are the data from Fig  Using the recombinase Cre to detect transfer of ICEBs1, irrespective of viability of transconjugants To better measure integration-mediated death of transconjugants, separate from the ability of transconjugants to form colonies, we developed a reporter system that uses Cre-mediated recombination to demonstrate transfer of ICEBs1. We reasoned that if ICEBs1 Δxis had integrated prematurely, then ICEBs1 genes must have been expressed. If the element encoded a heterologous recombinase (e.g., Cre), then transconjugants could be detected by virtue of a Cre-catalyzed recombination event, even if the transconjugants lost viability. Briefly, we inserted cre driven by Pxis in place of rapI-phrI in ICEBs1 (xis+, SAM830; Δxis, SAM892). Recipients (SAM599) contained a spectinomycin-resistance cassette (spc) with a promoter in the wrong orientation {(P(off)} between two lox sites. Transfer and expression of cre should cause inversion of the promoter-containing fragment such that the promoter would be driving expression of spc {P(on)} (S1 Fig). This Cre-mediated recombination is easily detectable by qPCR (S1 Fig and Methods) and does not require that the transconjugants be able to form colonies, although viable transconjugants should be spectinomycin resistant.
Cre-mediated recombination enabled us to compare the frequency with which recipients acquired ICEBs1 Δxis to that for ICEBs1 xis + . We mated ICEBs1 Δxis cre (SAM892) with recipients that contained the Cre reporter (SAM599). The acquisition efficiency as measured by CFUs was~0.01% transconjugants per recipient after two hours of mating (S2 Fig). In contrast, the acquisition efficiency as measured by Cre-mediated recombination (qPCR, Methods) was~5% per recipient after two hours of mating (Fig 2C), indicating that using Cre-mediated recombination, we detected 500-fold more inversions than viable transconjugants ( Fig 2D).
We found that the acquisition efficiency of wild-type ICEBs1 cre (xis + ) (SAM830) as measured by Cre-mediated recombination was~6% transconjugants per recipient after two hours of mating ( Fig 2C). This is virtually indistinguishable from the results with ICEBs1 Δxis (~5% transconjugants per recipient by the Cre assay). Based on this comparison and the results presented above, we conclude that ICEBs1 Δxis was transferred to recipients, that the element integrated prematurely into the genome of these transconjugants, and that the vast majority of transconjugants were not viable. We suspect that loss of viability was likely due to DNA damage from the integration of an element undergoing autonomous rolling circle replication (see below).

Premature integration of ICEBs1 in transconjugants causes chromosome abnormalities
To test if integration of ICEBs1 Δxis in naïve hosts caused effects on the host chromosome, we monitored chromosome segregation in transconjugants that had received ICEBs1 Δxis. We mated donors and recipients expressing different fluorescent proteins as described above (mApple in recipients; mNeongreen in donors and transconjugants) and then placed mating mixtures on agarose pads containing DAPI to visualize DNA (Methods). We found that~60% of transconjugants that received ICEBs1 Δxis had abnormal chromosome segregation, defined as any division event that did not result in symmetrical segregation of the chromosome to daughter cells, or the appearance of either compacted or elongated nucleoids. In contrast to the~60% for ICEBs1 Δxis, approximately 20% of ICEBs1 xis+ transconjugants had abnormal chromosome segregation (Fig 2E and 2F). These results indicate that premature integration of ICEBs1 in transconjugants caused an increase in chromosome abnormalities.
Detailed characterization of events in transconjugants is technically challenging due to the limited conjugation frequency and the need to distinguish transconjugants from the other two cell types (donors and recipients) present in the mating mix. Below, we describe experiments that overcome these limitations and allow for more detailed characterization of cells in which ICEBs1 is integrated and its genes are expressed.

Use of an ICEBs1 Δxis mutant host as a proxy for premature integration in transconjugants
Previous work indicated that activation of ICEBs1 that was unable to excise, due to integration in secondary sites, caused a host SOS response and cell death [23], similar to the phenotypes caused by premature integration of ICEBs1 Δxis in transconjugants. We used a strain (SAM249) containing ICEBs1 Δxis at its normal integration site (attB) as a proxy for premature integration of ICEBs1 Δxis in transconjugants. This is similar to the donor strain used in conjugation experiments described above, except without providing a functional xis in trans and with ICEBs1 Δxis integrated at its normal location in the chromosome (attB in trnS-leu2). Without any xis in the cell, activation of ICEBs1 Δxis results in de-repression of ICEBs1 gene expression and failure to excise from the chromosome, a situation similar to that caused by premature integration in the ICEBs1 Δxis transconjugants.
We found that one hour after activation (addition of xylose to induce Pxyl-rapI) of ICEBs1 Δxis in cells growing in defined minimal medium (Methods), cell viability was~2% of that before activation (Fig 3). In contrast, the number of viable cells of ICEBs1 xis+ increased 2-fold under the same conditions.

PLOS GENETICS
Timing of integration of a mobile genetic element is critical for fitness of the element and its host Activation of ICEBs1 Δxis caused cell death, apparently similar to that observed in ICEBs1 Δxis transconjugants. ICEBs1-containing cells were grown and treated as above and then placed cells on an agarose pad to track the fate of individual cells by time-lapse microscopy (S2 and S3 Movies). We found that during a three hour period following one hour of growth in ICE-inducing conditions, cell lysis occurred in~70% of cells containing ICEBs1 Δxis compared to~5% of those containing ICEBs1 xis+.
We found that the loss of viability of cells with ICEBs1 Δxis was due to loss of xis and not some unexpected effect on downstream genes or a cis-acting element in ICEBs1. We provided xis in trans from an ectopic locus (SAM388), thereby restoring excision. Viability was also restored to levels seen for cells with ICEBs1 xis+ (Fig 3), analogous to the restoration of the timing of integration observed when ICEBs1 Δxis transfers into recipients expressing xis (Table 1 and Fig 2A). Together, these results indicate that ICEBs1 Δxis is a suitable proxy for premature integration of ICEBs1 Δxis in transconjugants, that loss of cell viability is due to the inability of the element to excise from the chromosome, and that the terminal phenotype is cell lysis.

Cell death precedes lysis
Although cell lysis was the terminal phenotype, we found that cell death preceded lysis. We used staining with propidium iodide (PI) as an indicator of cell death (Methods; [24]) and tracked the ability of cells to take up PI and cell lysis using microscopy. We found that at the time ICEBs1 Δxis cells (SAM427) became PI-positive there was an appreciable decrease (~25%, n = 20) in cell surface area (S3A and S3B Fig), consistent with cell membrane permeability and loss of the proton motive force [25]. Cell lysis occurred on average within 12 minutes after staining with PI and the decrease in surface area. Together these results indicate that cell death caused by activation of ICEBs1 Δxis occurred before cell lysis. We suspect that loss of the proton motive force likely activates B. subtilis autolysins which then cause cell lysis [26,27].
The inability to excise does not immediately explain the loss of cell viability in ICEBs1 Δxis and indicates that there is another function of ICEBs1 biology that is lethal when the element is activated and unable to excise and when an active element is integrated prematurely in transconjugants.

Activation of ICEBs1 Δxis is lethal due to rolling circle replication of the element in the host chromosome
When activated, ICEBs1 undergoes unidirectional rolling circle replication that initiates from its origin of transfer (oriT) [8,28]. When the element is unable to excise due to deletion of the attachment site at the right end (ΔattR), this replication initiates in the integrated ICE and extends into the chromosome [8]. Replication of ICEBs1 requires three element-encoded functions: the relaxase NicK [15], the origin of transfer (oriT) that is nicked by the relaxase [15] and also functions as an origin of replication [8,28], and the helicase processivity factor HelP [28]. In an ICEBs1 ΔhelP mutant, nicking occurs at oriT [15], but the ICE DNA is not unwound and cannot undergo autonomous rolling circle replication [28].
We found that the lethality caused by activation of ICEBs1 Δxis was likely due to replication of the integrated element. There was no cell death in the ICEBs1 Δxis ΔhelP mutant (SAM393) (Fig 3), indicating that replication, or at least DNA unwinding, was required for cell death following activation of the excision-defective element. These results demonstrate that the loss of viability and subsequent lysis (above) caused by an activated ICEBs1 Δxis was most likely due to rolling circle replication of an element that is unable to excise. Further, cell lysis was the terminal phenotype, consistent with results above for early integration of ICEBs1 Δxis in transconjugants.

Activation of ICEBs1 Δxis induces the SOS response
Rolling circle replication of integrated ICEBs1 originates from oriT, extends into the chromosome, and generates ssDNA [15,28]. Iterative replication, that is multiple initiation events at the newly synthesized oriT, could also lead to a fragile region of the chromosome and potential arrest of chromosome replication forks that had initiated at the chromosomal origin, oriC. Both the increase in ssDNA and arrest or collapse of chromosomal replication forks could trigger the RecA-dependent SOS response.
We monitored induction of the SOS response in single cells using a transcriptional fusion of the reporter gene encoding the fluorescent protein mNeongreen to the promoter for yneA (yneA-mNeongreen), a gene that is repressed by LexA and activated during the SOS response {Methods and [29] and references therein}. We found that the SOS response was activated two hours after induction of ICEBs1 Δxis. The median fluorescence intensity of the population increased~20-fold compared to the uninduced ICEBs1 Δxis, with~83% of the population responsible for fluorescence (Fig 4A and 4B). In contrast, we found that two hours after activation of the excision competent ICEBs1 (xis+), the median fluorescence intensity of the population increased~2-fold compared to the uninduced ICEBs1 xis+, with~10% of the population responsible for fluorescence (Fig 4A and 4B). These results indicate that activation of ICEBs1 Δxis causes activation of the SOS response in the majority of cells in the population and are consistent with previous findings that ICEBs1 that is unable to excise from the chromosome causes induction of the SOS response [23].

Activation of ICEBs1 Δxis causes chromosome abnormalities
We found that activation of ICEBs1 Δxis caused chromosome abnormalities, analogous to those caused by premature integration of ICEBs1 Δxis in transconjugants. We grew cells containing ICEBs1 xis+ (SAM426) or ICEBs1 Δxis (SAM427) in defined minimal medium, activated ICEBs1 by overexpression of rapI (from Pxyl-rapI) for one hour, then removed the activator and placed cells on an agarose pad that supported rapid growth (doubling time of 20 minutes). To observe the nucleoid and formation of septa, agarose pads contained DAPI (DNA staining) and FM4-64 (membrane staining) and cells were visualized every three minutes for three hours (Methods). We found that the majority (~92%) of cells with an activated ICEBs1 Δxis had aberrant nucleoids and-or defects in chromosome segregation (Table 2 and Figs 4C and S4A-S4C, S4G and S4H and S4 Movie), including approximately 30% of cells with elongated nucleoids (S4D- S4F Fig). We defined aberrant chromosome segregation as nucleoids that had an asymmetric distribution between daughter cells. Daughter cells were defined as cells that existed for at least one frame after division and prior to lysis.
In addition to abnormal chromosome segregation and elongated nucleoids, many cells had defects in cell division, including mislocalized septa (septum formation not at mid-cell), guillotined chromosomes (S4D-S4F Fig), and anucleate cells (S4G and S4H Fig). We found that 29% of cells with an activated ICEBs1 Δxis had mislocalized septa, 6% had guillotined chromosomes, and 8% were anucleate. Additionally, approximately 65% of the population lysed during the course of the experiment ( Table 2).
Many more cells were affected by activation of ICEBs1 Δxis compared to activation of ICEBs1 that was able to excise (ICEBs1 xis+). Of the cells in which ICEBs1 (xis+) was activated and able to excise (SAM426), only~5% lysed, <15% had aberrant-looking chromosomes, and <10% had any apparent defect in cell division, compared to~65% lysed,~92% with chromosome abnormalities, and~43% with cell division defects in cells with ICEBs1 Δxis (Table 2).
Together, these results indicate that activation of ICEBs1 Δxis causes severe perturbations in chromosome organization and segregation, cell division, and ultimately causes cell death.

Chromosome abnormalities and cell lysis caused by activation of ICEBs1
Δxis were not due to activation of phage We found that the chromosome abnormalities and cell lysis caused by activation of ICEBs1 Δxis occurred in the absence of the functional phage SPß and the defective phage PBSX that are present in the B. subtilis chromosome. Both SPß and PBSX are activated during the SOS response {e.g., [29]} and cause cell lysis [30][31][32][33]. In a strain devoid of these two elements

PLOS GENETICS
Timing of integration of a mobile genetic element is critical for fitness of the element and its host (SAM520), activation of ICEBs1 Δxis still caused chromosome abnormalities and cell lysis ( Table 2 and S5 Movie). Based on these results, we conclude that activation of ICEBs1 Δxis causes chromosome abnormalities and cell lysis independently of these phage or phage-like elements.
There was always a fraction of cells in our experiments that had no detectable chromosome abnormalities, appeared to grow normally, and in contrast to most other cells in the population, did not lyse. This apparent normalcy could be because these cells were resistant to the effects of ICEBs1 Δxis, ICEBs1 Δxis was not activated, or the element had been lost. We used a fluorescent reporter fused to the major ICEBs1 promoter (Pxis-mApple; at an ectopic location in the chromosome; i.e., not in ICEBs1) that allowed us to distinguish between these possibilities. There is little to no fluorescence from Pxis-mApple in cells with ICEBs1 that is integrated in the chromosome and fully repressed (i.e., not activated) due to repression by the elementencoded repressor ImmR. Expression from Pxis is high in cells without ICEBs1 and intermediate in cells with ICEBs1 that had been activated (de-repressed) (S5 Fig).
We found that virtually all cells that had normal-looking chromosomes and that did not lyse had little or no detectable expression of Pxis-mApple (S6 Movie and S6 Fig), indicating that ImmR was present and fully active. Based on these results, we conclude that these cells still contained ICEBs1 Δxis, but that the element had not been activated. There were some cells that had activated ICEBs1 and had chromosome abnormalities but did not lyse (S6 Fig). It is possible that these cells were destined to lyse, but not during the three hours of our observations.

Discussion
We used a combination of population and single-cell analyses to study integration of ICEBs1 in transconjugants. We found that integration occurs several generations after initial transfer, early integration is lethal to most cells, and lethality is likely due to rolling circle replication from ICEBs1 that had integrated into the chromosome. Furthermore, based on our results, we conclude that premature integration of ICEBs1 in transconjugants causes phenotypes similar to those caused by activating an element that is unable to excise from the chromosome. In both cases, rolling circle replication from ICEBs1 in the chromosome causes a host SOS response, chromosome abnormalities, and cell death.
Multiple lines of evidence led to the conclusion that integration of ICEBs1 that is undergoing autonomous rolling circle replication causes host cell death. Most dramatically, activation of ICEBs1 that is unable to excise from the host genome caused cell death. Cell death was dependent on the element-encoded helicase processivity factor (HelP) that is needed for rolling circle replication (and DNA unwinding) of the element. In contrast, activation of ICEBs1 that is capable of excision (and autonomous rolling circle replication) did not cause cell death.

PLOS GENETICS
Timing of integration of a mobile genetic element is critical for fitness of the element and its host Further, premature integration of ICEBs1 in a transconjugant caused death of the transconjugant as determined by a drop in the number of viable transconjugants and as visualized by microscopy of single cells. Furthermore, there were chromosome abnormalities in cells in which ICEBs1 was integrated and undergoing autonomous rolling circle replication, but few chromosome abnormalities in cells in which ICEBs1 was integrated and not undergoing autonomous replicating, or in cells in which ICEBs1 was extrachromosomal and undergoing autonomous rolling circle replication. Together, our results demonstrate that robust viability and fitness of transconjugants depends on repression of replication before or concomitant with integration. Integration and ongoing rolling circle replication causes chromosome abnormalities and death of transconjugants. We conclude that ICEBs1 normally links integration, which is required for stable acquisition and maintenance of an element in a population [19], to the cessation of autonomous ICEBs1 replication. The ability of ICEBs1 to remain extrachromosomal and replicate autonomously is beneficial for the establishment and spread of the element, and for the hosts if the presence of the element confers a fitness advantage. Replication of ICEBs1 upon transfer to a new host allows for the element to segregate during growth and division of the transconjugants and for the transconjugants to become donors as soon as the ICE genes needed for conjugation are expressed. This is especially critical for transfer of the element in chains of cells and cells growing in communities [20,[34][35][36]. Below, we discuss the linkage between integration and cessation of autonomous replication and postulate that it is a conserved regulatory feature of integrative mobile genetic elements, including ICEs and temperate phages that integrate into the host genome.

Excisionases, required in donors, are also important in transconjugants
There is a large body of work demonstrating the importance of excisionases (recombination directionality factors) for the excision of ICEs and prophages from the chromosome of host cells [37]. However, it is largely assumed that excisionases are not important for establishment of an element after it enters a naïve host. We found that in addition to its well-known and expected role in excision from the chromosome [19], the ICEBs1 excisionase is critical following conjugative transfer of the element to a new host (transconjugant) and that it acts to affect the timing of stable integration.
Upon entry into a new host, most ICEBs1 genes are expressed as there is no repressor (ImmR) present and the major promoter Pxis is on (Fig 1C). This prevents integration and enables new transconjugants to serve as donors while the element is still extrachromosomal and expressing genes needed for autonomous rolling circle replication and conjugation, thereby increasing transfer of the element in a population [20,[34][35][36]. This de-repressed gene expression upon transfer to a naïve host is analogous to zygotic induction of phage lambda following its transfer to a new, non-lysogenic host via conjugation [38].
After transfer to a new host, ICEBs1 genes encoding the repressor (immR), anti-repressor (immA), and integrase (int) are also expressed (Fig 1C). Once enough ImmR accumulates, it activates its own promoter and represses transcription from Pxis [18,19]. This regulatory circuit also stimulates transcription of int. For ICEBs1, as with analogous systems, the presence of the excisionase directs the activity of the integrase so that excision is favored. Thus, integration of ICEBs1 would occur following the accumulation of the integrase and depletion of the excisionase. Depletion of the previously made excisionase occurs by dilution during cell growth and degradation if the protein is unstable. Although we have not yet tested it directly, we suspect that Xis is degraded by host proteases, thereby enabling integration shortly after repression of Pxis by ImmR, analogous to degradation of λ Xis [39].
Once repression of Pxis occurs, the ICEBs1 genes required for rolling circle replication are no longer expressed and autonomous replication will cease. In this way, integration of ICEBs1 into the chromosome of its new host is enabled, either concomitant with or shortly after the cessation of replication. This coordinated timing affects the fitness of ICEBs1 and host cells in several ways: 1) It allows replication and maintenance of ICEBs1 in the growing transconjugants prior to integration; 2) It allows early transconjugants to become donors before integration; and 3) It ensures that when the element does integrate into the genome, most of the integrants will be stable and viable with no autonomous replication of the integrated element.

Regulation and timing of integration of other ICEs
It is not known if premature integration of other ICEs causes phenotypes similar to those described here. However, there are several ICEs that are known to undergo autonomous replication, including Tn916 (Enterococcus faecalis) [9], ICEclc (Pseudomonas knackmussii B13) [7,10], ICEHin1056 (Haemophilus influenzae) [40,41], ICEMlSym R7 (Mesorhizobium loti R7A ) [16], SXT/R391 (Vibrio cholerae/Providencia rettgeri) [13], TnGBS1/TnGBS2 (Streptococcus agalactiae) [14], and ICESt1/ICESt3 (Streptococcus thermophilus) [42]. If autonomous replication of any of these ICEs occurs in transconjugants, then it seems likely that replication ceases prior to or concomitant with integration. We suspect that this is also true in the initial host if the element were to excise and then reintegrate. Additionally, some of these, including ICE-Hin1056 and TnGBS1/TnGBS2, are known to exist transiently as replicating elements in transconjugants, at least for a few generations prior to integration. These properties indicate that integration likely occurs after replication is halted.
ICEHin1056. Although it was not known to be an ICE at the time, prior work characterizing what is now known as ICEHin1056 revealed that donors, without detectable free plasmid during growth, were able to transfer antibiotic resistance to recipients. The resulting transconjugants transiently contained plasmid DNA and were capable of efficient transfer [40,41]. These initial studies indicated that at least some ICEs likely exist as extrachromosomal plasmids for several generations before integrating into the nascent host chromosome and that while plasmids, they were capable of efficient transfer to other cells.
TnGBS1/TnGBS2. ICEs in the TnGBS family differ from most characterized ICEs in that they use a DDE transposase rather than a phage-like integrase. The limited host range of TnGBS ICEs was found to be due to the inability of the element to replicate in species other than S. agalactiae. Further, qPCR analysis of the ratio of integrated ICE to circular ICE revealed that TnGBS ICEs remained in the circular form in transconjugants for some time before integration into the chromosome, and that the timing of integration varied between elements [14].
Other ICEs. For some ICEs, for example SXT/R391, ICEMlSym R7A , and ICEBs1, autonomous replication is critical for the maintenance of the element in donors following activation and excision [8,13,16]. In addition, for ICEBs1 and ICEMlSym R7A , replication is important in the generation of stable transconjugants [16,28]. Autonomous replication of ICEclc in donors increases the conjugation efficiency [7,10], and the element remains extrachromosomal in transconjugants for up to 10 hours after transfer [7]. For several ICEs, it has been shown that dysregulation between replication and integration results in instability of the element [8,9,13,16,19]. Together, the knowledge that various ICEs replicate autonomously indicates that proper regulation of integration is likely important for the stability of the element and survival of host cells.

Mechanisms for coupling integration with cessation of autonomous replication
There are at least three strategies used by ICEs to couple integration with cessation of autonomous replication: 1) Coordinated repression of excision and replication functions; 2) Excisionase-mediated activation of genes needed for replication; and 3) Physical separation of replication genes from their promoter upon integration.
1) Coordinated repression of excision and replication functions. ICEBs1 regulatory hierarchy involves the co-transcription of genes needed for excision, replication, and conjugation [8,[17][18][19]28]. The gene for the excisionase is upstream of those for replication and conjugation. As of yet, there are no published data on possible differential regulation of expression of excision and replication genes for ICEBs1. However, we note that a review of a transcriptomics study of B. subtilis [43] indicates an additional promoter downstream of the excisionase, driving expression of genes needed for replication and conjugation. The presence of this promoter could provide additional regulatory inputs that might affect the timing of integration and cessation of replication. Additionally, we suspect that the ICEBs1 excisionase (and probably others) is unstable, providing tighter temporal linkage between the repression of xis expression, disappearance of the protein, and the ability of the element to stably integrate.
2) Excisionase-mediated activation of genes needed for replication. ICEMlSym R7A appears to use a similar regulatory hierarchy as ICEBs1. Variants of ICEMlSym R7A are found in other Mesorhizobium species and have a similar regulatory hierarchy [44,45]. Additionally, for these ICEs the excisionase has been found to act as a transcription factor, activating the expression of downstream genes required for the transfer of the elements. This indicates that ICEs of the Mesorhizobium spp., may use a similar method of regulation observed for CTnDOT of Bacteroides thetaiotamicron, in which the excisionases moonlight as transcription factors, activating the expression of genes needed for conjugative replication, thereby directly linking Xis to the regulation of replication [45,46].
3) Physical separation of replication genes from their promoter upon integration. Tn916 of Enterococcus faecalis employs a regulatory hierarchy that involves physical separation of promoters from genes needed for replication and conjugation [47]. Excision of the element orients the promoter such that transfer and replication genes are transcribed. Conversely, integration separates the genes needed for replication and conjugation from the promoter(s) that drive their expression, thereby preventing further autonomous replication of the element.

Contrast between ICEs and conjugative plasmids
In contrast to ICEs, conjugative plasmids {reviewed in [4]} are typically maintained as extrachromosomal elements. However, conjugative plasmids can integrate into the host chromosome, typically by RecA-mediated homologous recombination between insertion sequences on the plasmid and the chromosome. For example, integration of the E. coli F plasmid into the chromosome generates an Hfr (high frequency of recombination) strain that can transfer chromosomal DNA to recipient cells [48]. Unlike ICEs, integration and excision of conjugative plasmids into the chromosome is not required for plasmid stability and is not viewed as a normal part of the plasmid lifecycle. In contrast to the largely uncontrolled replication from oriT of an integrated ICEBs1 that causes cell death, replication from oriT in an Hfr causes little or no cell death, but does cause a transient arrest of cell division [49]. This difference is likely due to limited nicking of and replication from oriT in the Hfr.

Conclusions
This linkage between integration and the absence of autonomous replication likely pertains to ICEs that undergo autonomous rolling circle replication and many temperate phages. Temperate phages that infect naïve hosts and integrate into the genome to form lysogens typically do so prior to the start of autonomous replication. This is typically achieved by a hierarchy of phage genes expressed at different times. Phage genes required for replication are typically expressed after the early genes and only if the phage enters the lytic cycle [50][51][52][53].
In summary, our work provides a new perspective on ICE regulation as it pertains to transconjugants. We uncovered a critical role for the excisionase in transconjugants and revealed the importance of regulatory mechanisms that delay integration until the cessation of replication. Because of the deleterious effects of integration before the cessation of replication, it is likely that most, and perhaps all integrative elements that undergo autonomous replication have a mechanism that enables integration only after replication is no longer possible. Because of the prevalence of ICEs, their powerful ability to transfer genes, and their recent uses in genetic analyses and engineering [54,55], it is important to understand ICE regulatory mechanisms and events in new hosts. This understanding could inform strategies to reduce the spread of undesirable elements (e.g, encoding clinically relevant antibiotic resistances) and facilitate the use of ICEs for engineering a wide range of microbes.

Media and bacterial growth
Cells were grown in LB medium or defined S7 50 minimal medium with 0.1% glutamate [56] and a carbon source, typically 1% L-arabinose, as indicated. Serial dilutions and resuspensions for plating viable CFUs were done in 1x Spizizen salts [57]. For experiments, indicated strains were colony purified from frozen (-80˚C) glycerol stocks on LB plates with the appropriate antibiotics. A single colony was picked and used to start a culture which was grown to midexponential phase. An aliquot was then diluted into fresh medium and allowed to grow to an appropriate culture density in mid-exponential phase for each experiment. Cells were typically grown in flasks at 37˚C with shaking (250 RPM).

Bacterial strains and alleles
All B. subtilis strains ( Table 3) are derivatives of JH642 (AG174; [59,60] and contain trpC and pheA mutations (not indicated in the table). Most strains were constructed by natural transformation to introduce appropriate alleles. In some cases, as indicated, ICEBs1 was transferred into strains by conjugation. ICEBs1 in most strains contained a gene conferring resistance to an antibiotic, often kan (kanamycin resistance), inserted in the rapI-phrI locus {e.g., Δ(rapI-phrI)342::kan} [17]. Strains cured of ICEBs1 [17] are indicated as ICEBs1 0 . Almost all alleles were first constructed in so-called 'clean' strains that were typically wild-type B. subtilis with  few if any other alleles. These were verified by phenotype, diagnostic PCR, and sometimes sequencing. Alleles were then transferred to other strains to build strains with complex genotypes. amyE::attB. SAM074 contains the chromosomal attachment site for ICEBs1 (attB) cloned into the nonessential gene amyE (amyE::attB). It is cured of ICEBs1, and contains a deletion of attB from its normal location (ΔattB::cat) that was described previously [23]. amyE::(attB spc) was made by cloning attB into the amyE integration vector pDG364 [62]. Oligonucleotides that contained the 60-bp attB were annealed and 25-bp of random sequence on the flanking ends were added and cloned into pDG364. The cat gene in pDG364 was then replaced with spc (spectinomycin resistance), that had been amplified from pMMB856 [58]. This plasmid (pSAM071) was then used to introduce attB into the B. subtilis chromosome at amyE by transformation into strain CAL522 (ΔattB::cat), selecting for resistance to spectinomycin.
Various versions of ICEBs1 were introduced into SAM074 (amyE::attB, ΔattB::cat) by conjugation with the relevant donor and SAM074 as recipient and selecting for transconjugants. Strains were verified by PCR and sequencing of relevant alleles. Strains constructed in this way include: SAM078; SAM207; SAM288; SAM472; SAM830; and SAM892. yddS::mls. yddS is~5 kb from attB and yddS::mls was used to distinguish recipients and transconjugants from donors, with mls serving as a counter-selective marker (MLS resistance) to prevent growth of donors, and a sequence for PCR that was present in only recipients and transconjugants. The allele contains an insertion of mls at base pair 22 of the 1,311 bp yddS open reading frame and was made by PCR amplifying~2 kb of DNA from upstream and downstream of the insertion site and assembling these (Gibson isothermal assembly [63]) such that they flanked a DNA fragment containing mls that had been amplified from pCAL215 [18]. This fragment was then cloned between the EcoR1 and KpnI sites in pUC19, generating pSAM095, which was then used to transform B. subtilis selecting for MLS resistance.
Ectopic expression of xis. Constructs for ectopic expression of ICEBs1 xis included cgeD:: (Pspank-xis) and cgeD::(Pxis-xis). Both were made with the plasmid pSAM178, a vector for integrating cloned DNA at cgeD. pSAM178 contains the gatB/yerO terminator adjacent to a multi-cloning site, kan (flanked by lox71 and lox66 sites for removal of kan by Cre-mediated recombination [64,65] as indicated, between sequences from cgeD in the pMMB124 [18] backbone. Plasmid pSAM189 contains Pspank-xis [19] in the pSAM178 backbone. pSAM276 contains Pxis-xis amplified from its endogenous location from 158 bp upstream of the xis open reading frame to 33 bp into ydzL, the gene downstream from xis, in the pSAM178 backbone. Both constructs were introduced into B. subtilis by natural transformation and selection for kanamycin resistance. Strains were confirmed by diagnostic PCR and checking for chloramphenicol sensitivity (loss of the plasmid backbone). kan was then deleted from these strains by introducing cre on a temperature sensitive plasmid that confers MLS resistance and expresses the Cre recombinase (pSAM097). Strains were then cured of pSAM097 by switching to growth All strains are derived from JH642 (AG174) [59,60] and contain the trpC2 and pheA1 alleles (not indicated in relevant genotypes). 2 The trnS-leu1-522 allele [23] should be present in SAM074 and all strains derived from it (SAM078, SAM207; SAM288; SAM472; SAM830; SAM892). For simplicity, this is not included in the relevant genotype. https://doi.org/10.1371/journal.pgen.1010524.t003 at a non-permissive temperature without selection. Strains with ectopic expression of xis include: SAM198, SAM207, SAM371, SAM379, SAM388, and SAM472. Fusions to mNeongreen, gfp, and mApple. amyE::PyneA-mNeongreen (in strains SAM426, SAM427) was used to monitor SOS induction. The fusion contains sequences extending from the first nucleotide to 151 bp upstream of the yneA open reading frame, including the promoter region [66,67]. This was fused to mNeongreen and flanked by sequences for recombination into amyE, including spc, by linear Gibson isothermal assembly [63]. The resulting construct was then transformed into B. subtilis selecting for spectinomycin resistance to generate strain SAM352. Genomic DNA from SAM352 was used to move amyE::{(PyneA-mNeongreen) spc} into various strains by transformation selecting for spectinomycin resistance.
Ppen-mNeongreen in ICEBs1 (ICEBs1 Δ(rapI-phrI)342::{(Ppen-mNeongreen) kan} in SAM268; SAM288; SAM472) was used to visualize ICEBs1 in single cells. The constitutive promoter Ppen was amplified from pMMB1010 [20] (with a spoVG ribosome binding site in one of the PCR primers downstream of Ppen for efficient translation of mNeongreen) and fused to mNeongreen. This fusion was flanked by sequences upstream and downstream from the endpoints of Δ(rapI-phrI)342 and included kan [17] using Gibson isothermal assembly. This was then transformed into B. subtilis (ICEBs1+), selecting for kanamycin resistance and verifying that the reporter was in ICEBs1.
amyE::Ppen-mApple (SAM271) was used to distinguish recipients and transconjugants from donors in conjugation experiments visualizing single cells. The constitutive promoter Ppen was amplified from pMMB1010 [20], with the ribosome binding site from spoVG (as above), and placed upstream of mApple. This fragment was then cloned into the amyE integration vector pDR160 [21], replacing the xylose-inducible cassette, followed by integration of Ppen-mApple at amyE by double-crossover homologous recombination (between the amyE fragments in pDR160) selecting for spectinomycin resistance.
amyE::Pxis-gfp (in SAM005, SAM011, and SAM032) and amyE::Pxis-mApple (in SAM168, SAM456) were used to report on expression levels of Pxis, as a reflection of the amount of active ImmR (the ICEBs1 repressor) in the cells. Pxis was amplified from its endogenous location from 215 bp to 4 bp upstream of the xis open reading frame, fused to gfp or mApple, and cloned into the amyE-integration vector pDG364 [62], generating plasmids pSAM003 (Pxisgfp) and pSAM159 (Pxis-mApple). These plasmids were transformed into AG174 or JMA222 by natural transformation selecting for chloramphenicol resistance generating strains SAM005 (Pxis-gfp) and SAM168 (Pxis-mApple). Strains were confirmed by diagnostic PCR and sequencing. Genomic DNA was used to move alleles by transformation to build other strains, including SAM011 and SAM456.
ΔSPß::(lox-kan-lox), ΔPBSX::(lox-cat-lox), Δskin::(lox-spc-lox) in strains SAM520 and SAM529. Deletion-insertion mutations of the prophage SPß and the defective phages PBSX were described [61]. These essentially delete the indicated phage and insert an antibiotic resistance gene flanked by lox sites. The Δskin::(lox-spc-lox) mutation was constructed similarly, by PCR amplifying genomic DNA from upstream and downstream of skin and then using Gibson assembly to place these PCR products fragments on either side of lox-spc-lox. This was then transformed into B. subtilis, selecting for spectinomycin resistance, resulting in deletion of skin and insertion of lox-spc-lox. Genomic DNA was used to move alleles by transformation to build strains SAM520 and SAM529. Where indicated, the antibiotic resistance gene was deleted by Cre-mediated recombination as above.
ICEBs1 Δ(rapI-phrI)342::Pxis-cre kan in strains SAM828, SAM830, and SAM892. cre, encoding the Cre recombinase, was cloned into ICEBs1 and used to monitor transfer of the element to cells that might otherwise not remain viable. Transconjugants could be detected with a reporter that would recombine in the presence of Cre. To construct Pxis-cre in ICEBs1, the promoter Pxis was amplified from its endogenous location from 158 bp upstream of and to 4 bp into the xis open reading frame. cre was amplified from pDR244 [68] and the ribosome binding site from spoVG [69] was contained on one of the primers and placed upstream of cre to increase translation efficiency. Two DNA fragments, one upstream and one downstream of the endpoints of Δ(rapI-phrI)342, were amplified, including kan, and the four fragments were assembled using linear Gibson isothermal assembly. This was then introduced into ICEBs1 by transformation selecting for kanamycin resistance.
The reporter in recipient cells (SAM599) that was used to indicate acquisition of cre and subsequent Cre recombinase activity (S1 Fig) was constructed by Emily Bean (original strain ELC69). It consists of a promoter flanked by lox sites [64] such that recombination would result in a stable inversion [65]. The lox71-promoter-lox66 fragment was upstream from spc. The initial orientation is such that the promoter is pointed away from spc {P(off)}. Cre-mediated recombination would invert the promoter-containing fragment such that the promoter would be driving expression of spc {P(on)}. Viable transformants that underwent Cre-mediated recombination would be spectinomycin resistant. Recombinants are also easily detectable by PCR with primers diagnostic of the inversion (S1 Fig). This assay does not require cell viability. Strain SAM698 contains the inversion {P(on)} in all cells and was used as a standard for qPCR.
Cre-mediated inversion appeared to underestimate the conjugation efficiency by approximately three-fold. In a direct comparison, the acquisition efficiency of ICEBs1 was~20% transconjugants per recipient after two hours of mating, as measured by CFUs (S2 Fig) compared to~6% as measured by Cre-mediated recombination (Fig 2C). This discrepancy decreased at later times and after three hours of mating,~82% of the transconjugants (Fig 2D) had a detectable inversion based on the Cre reporter. Virtually every viable transconjugant (>99.8%) recovered as a colony was resistant to spectinomycin, indicating that Cre catalyzed an inversion in virtually all transconjugants, but that there was a delay between the initial acquisition of ICEBs1 (as measured by CFUs) and the Cre-catalyzed recombination (inversion) event.

Mating assays
Cells were grown in 25 ml flasks in LB medium in a water bath at 37˚C and 250 RPM shaking until they reached an OD 600 of~0.12. Arabinose and xylose (1% each) were added to donor strains to induce expression of Pxyl-rapI, thereby causing cleavage of the ICEBs1 repressor ImmR [18,22] and de-repression of ICEBs1 gene expression. Where appropriate IPTG (2 μM) was added to induce expression of genes needed for complementation. Cultures were grown for 40 minutes after de-repression of Pxyl-rapI to an OD600 of~0.3-0.45.
Donors and recipients (in mid-exponential growth) were mixed at ratio of 2:1 (resulting in the equivalent of 3 OD600 units), vacuum-filtered, and filters were transferred to Spizizen minimal salts plates as described previously [17]. We found that these conditions resulted in high conjugation efficiencies that enabled detection of ICEBs1 integration by qPCR. Mating reactions were allowed to go for various amounts of time, as indicated, and were halted by placing filters in 3 ml of Spizizen minimal salts and vigorously vortexing. An aliquot of this mating mix was then used for serial dilutions and plating for colony forming units with appropriate antibiotics for selection of donors, recipients, and trasnsconjugants. The bulk of the mating mix was harvested for extraction of genomic DNA.

Quantitative PCR assays to measure integration of ICEBs1 and Cremediated recombination
Quantitative PCR was used to measure integration of ICEBs1 in transconjugants. Cells were lysed with 40 mg/ml of lysozyme and genomic DNA was isolated using Qiagen DNeasy kit. qPCR was done using SSoAdvanced SYBR master mix and CFX96 Touch Real-Time PCR detection system (Bio-Rad).
Serial dilutions were prepared from genomic DNA for each sample. The averages of at least two dilutions with three technical replicates each within the linear range were used to determine relative amount of ICEBs1 integration or Cre-mediated recombination by the Δct method. These were normalized to CFUs, as indicated, to determine the fraction of the recipient or transconjugant population with each marker.

Microscopy and image acquisition
We used an inverted microscope (Nikon, Ti-E) with a motorized stage placed in a temperature-controlled box (In Vivo Scientific) at 37˚C. Images were acquired with a CoolSnap HQ2 camera (Photometrics). Fluorescence was generated using a Nikon Intensilight mercury illuminator through appropriate sets of excitation and emission filters (filter set 49000 for DAPI, 49003 for mNeongreen, and 49008 for mApple; Chroma). Samples were spotted onto agarose pads that were set in a homemade incubation chamber made by stacking two sealable Frame-Seal Incubation Chambers (BIO-RAD). Agarose pads contained 1.5% Ultrapure agarose (Thermo Fisher) dissolved in either minimal medium with Spizizen salts or CH medium [57]. Minimal medium was used when assessing fluorescence intensity and CH medium was used for tracking cell viability. For time-lapse experiments, agarose pads were supplemented with the appropriate carbon source at 1% L-arabinose, 1% glucose, and 1% D-xylose w/v. Where applicable, agarose pads were supplemented with 25 ng/ml DAPI, 170 ng/ml FM4-64, and 0.1M propidium iodide. For visualization of cell membranes, cultures were incubated with FM4-64 (170 ng/ml) for 1 hour before imaging.

Microscopy image analysis
Images and videos were processed with Fiji (ImageJ) [70]. All images for fluorescent channels were subjected to background subtraction using a rolling ball radius of 10 or 20 pixels (10 pixels was used for images with foci). Cell meshes were acquired using Oufti [71]. Meshes were analyzed in MATLAB using custom scripts. The blue triangle in recipients and transconjugants indicates the reporter for and result of Cre-mediated inversion. Blue and yellow triangles indicate lox sites as labeled. The region between the blue and yellow triangles contains a promoter, either driving transcription of spc (With Cre; bottom right) or in the opposite orientation (No Cre; top). Blue block arrow represents spc open reading frame. Inversion in the presence of Cre produces lox71/lox66 and loxP sites which are incapable of subsequent recombination [64]. The horizontal black arrows with letters A, B, and C represent primers for PCR to detect the reporter and Cre-mediated inversion. PCR with primers B and C detect the product of Cre-mediated recombination. PCR with primers B and A detect the starting reporter with the promoter in the 'off' orientation. Cells with the reporter Pxis-gfp were grown at 37˚C in a defined minimal medium with L-arabinose. Expression of rapI (Pxyl-rapI) was induced by the addition of D-xylose. After 2 hrs cells were spotted onto agarose pads and imaged for fluorescence. Each data point represents the fluorescent intensity from a single cell. The longest horizontal line represents the mean and bars above and below represent the standard deviation. No reporter is the parental strain with no genetic markers (AG174). ICEBs1 0 is a strain cured of ICEBs1 (SAM011) and expression from Pxis is in the absence of the repressor ImmR. ICEBs1 OFF is a strain containing ICEBs1 Δ(rapI-phrI)::kan (SAM032) without exogenous activation of the element. These cells produce active ImmR and transcription from Pxis is repressed and there is only basal expression of ICEBs1. ICEBs1 ON is the same strain (SAM032) except that ICE has been activated by virtue of expression of rapI and subsequent inactivation of the repressor ImmR. Pxis is expressed under these conditions. (TIF)

S6 Fig. Cells that continue to grow normally following activation of ICEBs1
Δxis in the bulk culture still have the element. Cells containing ICEBs1 Δxis and the reporter Pxis-mApple (SAM456) were grown at 37˚C in a defined minimal medium with L-arabinose. Expression of rapI (Pxyl-rapI) was induced by the addition of D-xylose. After 60 mins cells were washed and spotted onto agarose pads with CH medium and DAPI. The image shown is after 3 hrs on the agarose pad and is a composite of the red fluorescence (Pxis-mApple) and DAPI channels. Red fluorescence is pseudo-colored red and DAPI is pseudo-colored blue. The white arrow points to chain of normally growing cells that have ICEBs1 but the element has not been activated. Had they lost ICEBs1, or had ICEBs1 been activated, then Pxis-mApple would be expressed. (TIF) S1 Movie. The fate of ICEBs1 Δxis transconjugants. Donors (ICEBs1 Δxis, SAM472) harboring ICEBs1 with constitutively expressed mNeongreen were mated with recipients (SAM271) constitutively expressing mApple for 4 h. Following mating, samples were spotted onto nutrient-rich agarose pads containing erythromycin and lincomycin (recipients and transconjugants contain mls and are resistant to these two antibiotics). Donors appear pale yellow. Recipients appear red. Transconjugants are both red and yellow and appear orange or bright yellow. (AVI) S2 Movie. Activation of ICEBs1 Δxis (unable to excise) causes cell lysis. Cells (SAM427) were grown at 37˚C in a defined minimal medium with L-arabinose. Expression of rapI (Pxyl-rapI) was induced by the addition of D-xylose. After 60 mins cells were washed and spotted onto to nutrient-rich agarose pads and imaged every 3 mins for 3 hrs. (AVI) S3 Movie. Activation of WT ICEBs1 does not cause cell lysis. Cells (SAM426) were grown at 37˚C in a defined minimal medium with L-arabinose. Expression of rapI (Pxyl-rapI) was induced by the addition of D-xylose. After 60 mins cells were washed and spotted onto nutrient-rich agarose pads and imaged every 3 mins for 3 hrs. (AVI) S4 Movie. Activation of ICEBs1 unable to excise causes chromosome abnormalities. Cells (SAM427) were grown at 37˚C in a defined minimal medium with L-arabinose. Expression of rapI (Pxyl-rapI) was induced by the addition of D-xylose. After 60 mins, cells were washed and seeded to nutrient-rich agarose pads containing DAPI and imaged every 15 mins for 45 mins and then every 3 mins for 2.25 hrs. (AVI) S5 Movie. Chromosome abnormalities caused by activation of ICEBs1 that is unable to excise are not due to SPβ, PBSX, or skin. Cells (SAM529) were grown at 37˚C in a defined minimal medium with L-arabinose. Expression of rapI (Pxyl-rapI) was induced by the addition of D-xylose. After 60 mins cells were washed and spotted onto nutrient-rich agarose pads containing DAPI and imaged every 15 mins for 45 mins and then every 3 mins for 2.25 hrs. (AVI) S6 Movie. Cells containing ICEBs1 Δxis that survive induction are those that have not activated the element. Cells (SAM456) were grown at 37˚C in a defined minimal medium with Larabinose medium. Expression of rapI (Pxyl-rapI) was induced by the addition of D-xylose. After 60 mins cells were washed and spotted onto nutrient-rich agarose pads containing DAPI and imaged every 15 mins for 45 mins and then every 3 mins for 2.25 hrs. (MP4) S1 Data. Underlying raw data for experiments presented. The excel spreadsheet contains the underlying data for the experiments presented in tables and figures. (XLSX)